The proposed longitudinal FDG-PET investigation is designed to identify the earliest predictors of memory and brain deterioration in preclinical Alzheimer's disease (AD) and to examine a potential mechanism for this early damage. Four important new observations converge to support our aims: 1. longitudinal FDG-PET and MRI studies show reductions at baseline in the entorhinal cortex (EC) rate of glucose metabolism (MRglu) and size in normal subjects (NL) who decline to mild cognitive impairment (MCI), a known risk factor for AD; 2. reduced EC MRgIu and size changes predict over 3 years the appearance of hippocampal deficits in subjects that declined to MCI; the MRglu of the EC and hippocampus obtained using MRI-registered and atrophy- corrected procedures are superior to MRI volumes in the classification of NL, MCI and AD; 4. in vivo, AD patients show a reduction in the ratio of glucose to 02 consumed and a reduction of the hippocampal MRgIu response to cortisol, glucose, or cognitive challenge. Together these observations show that the earliest known brain lesions of AD are in the hippocampal formation (HipForm) and suggest a physiological defect in the regulation of glucose metabolism. We will conduct over 5 years a 3-year follow-up study of NL subjects. Aim 1: to evaluate the EC MRglu as a predictor of MCI. We will test two hypotheses: 1) EC MRgIu reductions predict progressive declarative memory decline and progressive hippocampal and neocortical metabolism and volume reductions, and 2) EC MRglu reductions are anatomically specific predictors of progressive memory impairment. Three groups (each n=35) will be studied. Group 1: healthy NL 50-70 yr old subjects with longitudinally observed memory decline; Group 2 a demographically matched NL group without a history of memory decline; and Group 3,20-40 year old NL subjects studied once for use as a reference. Over 36 months, Groups 1 and 2 will receive 3 clinical evaluations with MRI. FDG-PET will be obtained twice, at baseline and endpoint. Aim 2: to evaluate the effects of acute steady state hyperglycemia vs. saline on memory performance and on HipFom, MRglu and kinetics (dynamic PET scan). Fifteen subjects from each of groups 1 and 2 will be selected. We will test two hypotheses: 1) under acute hyperglycemia, Group 1 will show deficits in HipForm MRglu and kinetics and will not enhance memory function, and 2) diminished HipForm glucose transport predicts longitudinal brain and memory changes. We have strong preliminary evidence in support of our hypotheses and all the facilities are in place.